Twitterverse Debate Visualizations

November 5, 2016

David Leblang (Department of Politics Chair) and Paul Freedman (Associate Chair) working with Hope McIntyre and Yifeng Song, of the Data Science Institute at the University of Virginia developed the following three visualization tools to analyze Twitter activity. Initially the tools will be used to analyze tweets during presidential debates on Monday, September 26, 2016, Sunday, October 9, 2016, Wednesday, October 19, 2016, and the vice presidential debate on Tuesday, October 4, 2016. Coverage will continue following the election on Tuesday, November 8.

We recommend using the most current browser available. Older browsers and operating systems may not have complete emoji font sets and older browsers don’t support emojis at all.

Emoji Directed Tweets

The Emoji Analysis tracks tweets which have emojis attached to the message and are directed to particular candidates using the @ sign (e.g. @HillaryClinton and  @realDonaldTrump for the presidential debates, and @timkaine and @mike_pense for the vice-presidential debate). Tweets tagged with the @ symbol are a way for the sender to direct their message to the candidate; these tweets are received by the candidate (though not necessarily read).

There are over 900,000 different emojis, though many of these could just be considered variations by brand (Apple iPhones and computers, Google Android devices, Windows devices, and other, as well as emoji versions customized by the user), with slightly different images.

(Updated 10-24-16, includes final debate data)

Hashtag Visualization

The Hashtag Visualization allows the user to click on graph traces to reveal hashtag phrases (e.g. #imwithher, #makeamericagreatagain). These tweets are also directed to the candidate using the @ symbol. Hashtags allow the Twitter user to categorize a tweet, revealing or explaining intent. Hashtags are useful for responses to events (in this case the presidential debates, but also for other events such as earthquakes and celebrity meltdowns). Hashtags frequently are used sarcastically.

(Updated 10-24-16, includes final debate data)

Directed Tweets Visualization

The Directed Tweets by Volume shows the number of Tweets directed to a candidate over the duration of a day. Count is reset at midnight.

(Updated 10-24-16, includes final debate data)

Twitter Follower Data

This graph shows the growth in Twitter followers by candidate during the debate, by minute. It shows that Clinton saw significantly greater growth in followership than Trump: During the course of the debate she gained almost 58,000 new Twitter followers, while Trump added less than 36,000. Importantly, Clinton’s advantage continued even after the debate, while pundits and commentators discussed the relative performance of the two candidates, and the picture of a Clinton “win” began to solidify. By 11PM (10:51), Clinton had added almost 71,000 followers in total, while Trump gained only 42,000 new followers.
It is important to note, however, that as the candidates headed into the debates, Trump enjoyed a huge advantage in Twitter followers, with 11.7 million, versus fewer than 8.9 million for Clinton. By the end of the evening, although Trump was still ahead by more than 2.7 million followers, Clinton’s s had narrowed the gap every so slightly.

New Content Coming Soon!